It has become clear that there is no single type of breast cancer, but rather at least six sub-types of human cancer have been identified through gene expression profiling. Our recent work has focused on understanding which mouse models of human mammary cancer represent specific sub-types of human breast cancer in order 1) to better understand what determines the mechanisms that underlie the lineage determinants of breast cancer sub-types, and 2) which models best represent particular sub-types of human breast cancer for use in pre-clinical testing for those sub-types. Recently, we have identified in mouse models where functions of both p53 and Rb have been compromised a very comprehensive, integrated genetic network of genes related to replication, DNA synthesis and repair, chromosome maintenance, proliferation, and apoptosis. This network is highly represented in- and predictive of basal-type breast cancer and the most aggressive forms of prostate and lung cancers. We hypothesize that many of these previously unrecognized genes related to basal type tumors may be potentially important targets for anti-cancer therapies and we are pursuing this experimentally. To this end, we have performed many in vitro and in vivo studies using single drugs or combination therapies with promising results. Rational combinations were chosen based upon the genetic pathways shown to be efficacious using our siRNA screen. More recently, we have performed synthetic lethal screens using taxanes, gemcitabine and CHK1 inhibitors in combination with our focused siRNA library and have identified potentially useful combination therapies which are being tested in our in vivo model systems. We have also expanded our work related to understanding global genomic changes that occur in GEM mammary tumor models based upon the initiating oncogenic event. We have developed gene expression, array CGH and miRNA datasets from the same tumors from eight mammary tumor models. We hypothesize that cancer may evolve differently depending upon the initiating oncogenic event and that understanding these processes may provide insights into identifying secondary changes that are critical for tumor development and progression. We have determined that models with compromised function of p53, Rb or BRCA1 evolve into a basal-type of mammary cancer, whereas the MMTV-oncogene driven models reflect a luminal phenotype. Interestingly, significant differences in genomic copy number changes and ploidy also distinguish these various mammary tumor models. Analysis of miRNA expression patterns also differ significantly between the models and the functional significance of this is being explored and compared to miRNA expression patterns in human breast cancer. Many of these distinguishing molecular features may be related to the lineage specificity of the mammary tumors that develop. We have recently used a novel informatics approach to correlate inverse relationships between miRNA and mRNA expression of predicted miRNA targets to identify new miRNA targets that may be related to tumor lineage and biology. We have continued to characterize several new mouse models of mammary cancer that demonstrate the critical interactions of genetic mutations in altering the histologic tumor phenotype that develops. Preliminary analyses indicate that the loss of expression of a particular gene can switch the tumor lineage from an adenosquamous to adencarcinoma phenotype. The molecular mechanisms related to this lineage specificity are being further explored through microarray analyses and functional studies. We have also begun to demonstrate that targeted expression of key transcription factors to breast cancer cells can significantly alter their phenotype and biologic properties. We are pursuing this approach as a means of novel differentiation therapy with the goal of reducing aggressive and metastatic properties of breast cancer cells. Ultimately, this could have translational value for prevention or therapeutic approaches. It has become clear that tumor cells do not exist as isolated entities, but interact in critical ways with the microenvironment and neighboring stromal cells. We are pursuing studies to explore cross talk between tumor epithelial cells and stromal cells under various conditions using genomic technologies. This will expand our understanding of systems biology where information exchange in a multicellular system will be determined. This will likely have important implications for identifying potential targets to interrupt this important cross-talk between tumor cells and their environment. During this year, important collaborations with Dr. Joe Gray at Lawrence Berkeley National Labs were initiated to further identify similarities between human breast cancer and mouse mammary tumor models using gene expression arrays, miRNA arrays, and array CGH. An important goal of this work is to identify potential therapeutic targets, particularly for triple-negative breast cancer. Additionally,I have establish an exciting collaboration with Dr. Laura V'ant Veer at UCSF through the I-SPYII clinical trial. This will allow us to perform preclinical studies using mouse models as a means of determining the predictive value of models. Further, new combination therapies will be tested in pre-clinical models that could potentially be used in the I-SPYII trial.